silu(gate(x))·up(x) for 2..4 activation rows sharing each decoded weight (variants 7..13), gate and up at the same width. Borrows x and the weights; returns owned lazy output with x's leading dimensions.
silu(gate(x))·up(x) for 2..4 activation rows sharing each decoded weight (variants 7..13), gate and up at the same width. Borrows x and the weights; returns owned lazy output with x's leading dimensions.